🔀 Tutorial 3: Network Core – Packet Switching and Circuit Switching

COMP347 (Revision 10) | TrustOpen University

🎯 Learning Objectives

Upon completion of this expanded tutorial, students will be able to:

• Explain the functions of the network core: forwarding, routing, and switching.
• Compare packet switching and circuit switching using quantitative models of delay and utilisation.
• Describe the store‑and‑forward mechanism and calculate transmission delays.
• Define statistical multiplexing and contrast it with FDM and TDM.
• Analyse the performance trade‑offs between packet and circuit switching in different traffic scenarios.
• Explain the hierarchical ISP structure and the roles of peering, transit, and IXPs.
• Understand the concept of virtual circuits and their use in MPLS.
• Apply basic queuing theory (M/M/1) to estimate queuing delay.

🔭 Overview

The network core is the backbone of the Internet—a vast interconnected mesh of routers and high‑speed links that moves data across continents. This tutorial provides a rigorous, mathematically‑grounded examination of the core's two fundamental switching paradigms: packet switching and circuit switching. We dissect the store‑and‑forward operation, statistical multiplexing, and the queuing dynamics that lead to delay and loss. We then analyse circuit switching with its resource reservation and deterministic performance. Through quantitative examples and comparison tables, we illuminate why packet switching dominates modern data networks. We also explore the Internet's hierarchical structure, peering and transit agreements, and the hybrid approach of virtual circuits. Finally, we introduce queuing theory models that enable engineers to dimension buffers and predict performance.

1. The Network Core: An Overview

The network core comprises the interconnected packet switches (routers) and communication links that form the Internet's backbone. Its primary functions are:

• Forwarding: moving packets from an input link to the appropriate output link at each router.
• Routing: determining the end‑to‑end path using routing protocols (e.g., OSPF, BGP).
• Switching: transferring data from one link to another using either packet or circuit switching.

Key components: high‑speed fibre links (10–400 Gbps), routers with large forwarding tables, and Internet Exchange Points (IXPs) where ISPs interconnect.

2. Packet Switching – Principles and Mechanics

2.1 Store‑and‑Forward and Transmission Delay

In packet switching, data is divided into packets. Each router must receive the entire packet before forwarding (store‑and‑forward). This allows error checking and routing decisions.

Transmission delay at a link: \( d_{trans} = L / R \), where \( L \) is packet length (bits) and \( R \) is link rate (bps). For a path with \( N \) links, the total transmission delay (excluding propagation and processing) is \( N \cdot L / R \).

2.2 Statistical Multiplexing

Statistical multiplexing dynamically allocates bandwidth to users based on their actual traffic demand. Unlike FDM/TDM, where bandwidth is fixed, statistical multiplexing achieves high utilisation because idle users do not consume capacity. The aggregate traffic can exceed the link rate temporarily, but over time the average is within capacity, leveraging the bursty nature of data traffic.

2.3 Queuing and Loss

Each output link has a queue (buffer). If the average arrival rate exceeds the service rate, the queue grows; when full, packets are dropped (loss). The traffic intensity \( \rho = \frac{L \cdot a}{R} \) (where \( a \) is arrival rate in packets/s) determines queuing behaviour:

• \( \rho < 1 \): queue is stable; queuing delay is bounded.
• \( \rho = 1 \): queue grows without bound.
• \( \rho > 1 \): queue length increases indefinitely; loss occurs.

2.4 Advantages and Disadvantages

Advantages: efficiency, resilience (rerouting), scalability, flexibility. Disadvantages: variable delay, packet loss, no guaranteed QoS without additional mechanisms.

3. Circuit Switching – Dedicated Communication

3.1 Phases of Operation

1. Circuit establishment: resources are reserved along the path.
2. Data transfer: data is transmitted over the dedicated circuit.
3. Circuit teardown: resources are released.

3.2 Multiplexing in Circuit Switching: FDM and TDM

• FDM (Frequency Division Multiplexing): each circuit gets a dedicated frequency band.
• TDM (Time Division Multiplexing): each circuit gets a dedicated time slot in a repeating frame.

Both allocate fixed capacity, leading to inefficiency during silent periods.

3.3 Advantages and Disadvantages

Advantages: guaranteed bandwidth, fixed delay (no queuing), simple forwarding. Disadvantages: inefficiency, setup overhead, less resilient.

4. Packet vs. Circuit Switching: A Quantitative Comparison

4.1 Performance Analysis Example

Consider a 1 Mbps link. Each user generates 100 kbps when active, and is active 10% of the time.

• Circuit switching: can support at most 10 users (10 × 100 kbps = 1 Mbps). Additional users are blocked.
• Packet switching: with 35 users, the probability of more than 10 active simultaneously is very small (binomial distribution). Average throughput = 35 × 0.1 × 100 kbps = 350 kbps. The network can accommodate occasional bursts.

This illustrates the statistical multiplexing gain of packet switching.

4.2 Comparison Table

5. Internet Structure: A Network of Networks

5.1 ISP Hierarchy

• Tier‑1: global backbone (e.g., AT&T, Verizon, Deutsche Telekom). Interconnect via settlement‑free peering.
• Tier‑2: regional/national, connect to Tier‑1 and peer with others.
• Tier‑3: local access providers, connect to Tier‑2.

5.2 Peering, Transit, and IXPs

• Peering: direct traffic exchange, often free.
• Transit: paid access to the wider Internet.
• IXPs: physical locations where multiple ISPs interconnect (e.g., DE‑CIX, AMS‑IX).

6. Virtual Circuits and MPLS

Virtual circuit switching combines aspects of packet and circuit switching. A setup phase establishes a path and assigns a virtual circuit identifier (VCI). Each packet carries the VCI, and routers forward based on it (no longest prefix match). This provides fast forwarding and can reserve resources.

MPLS (Multi‑Protocol Label Switching) is a widely used virtual‑circuit technology in ISP backbones for traffic engineering and VPNs. It uses labels (short, fixed‑length) for fast forwarding.

7. Queuing Theory and Delay Models

The M/M/1 queue (Poisson arrivals, exponential service times, single server) provides closed‑form expressions:

• Traffic intensity: \( \rho = \lambda / \mu \), where \( \lambda \) is arrival rate, \( \mu \) is service rate.
• Average number in queue: \( N_q = \rho^2 / (1 - \rho) \).
• Average queuing delay: \( W = \rho / (\mu - \lambda) \).

For finite buffers (M/M/1/K), loss probability is given by \( P_{loss} = \frac{(1-\rho)\rho^K}{1-\rho^{K+1}} \) (for \( \rho \neq 1 \)).

These models are essential for buffer sizing and QoS provisioning.

📝 Quiz: Tutorial 3

✏️ Exercises: Tutorial 3

📚 Homework: Tutorial 3

📌 Summary

This tutorial has explored the network core in depth:

• The core functions: forwarding, routing, and switching.
• Packet switching: store‑and‑forward, statistical multiplexing, queuing, and loss; advantages in efficiency and resilience.
• Circuit switching: dedicated paths, FDM/TDM, guaranteed performance; drawbacks in efficiency and setup overhead.
• A quantitative comparison showing the statistical multiplexing gain of packet switching.
• The Internet's hierarchical structure with ISPs, peering, transit, and IXPs.
• Virtual circuits and MPLS as hybrid approaches.
• Queuing theory (M/M/1) to model delay and loss.

Understanding the network core is essential for grasping how data moves across the globe. In the next tutorial, we will delve into performance metrics: delay, loss, and throughput, and their impact on applications.